IRC channel logs
2021-09-10.log
back to list of logs
<damo22>Pellescours: it should not be attempting to build libc from netbsd <damo22> HOST_CC=gcc LDADD="-L/usr/lib/i386-gnu -L/usr/lib/gcc/i686-gnu/10" _GCC_CRTENDS= _GCC_CRTBEGINS= _GCC_CRTI= _GCC_CRTN= CFLAGS=-Wno-error=stack-protector ./build.sh -V TOOLS_BUILDRUMP=yes -V MKGROFF=no -U -u -T ../tools -m i386 tools rump <damo22>also, you need netbsd-src from upstream <damo22>there are 4 patches, 3 are merged <damo22>i just did a clean build, with an extra LDADD library path and it succeeded <damo22>it needs -L/lib/i386-gnu as well <damo22>Pellescours: I built this commit from upstream, plus my last patch: <damo22>* c29caa2cb (HEAD -> test-rump-upstream) build.sh: Wrap host toolchain for rump builds <damo22>* 904baec69 (netbsd/trunk) lint: remove array creation from expr_derive_type <Pellescours>damo22: I’m on fresh hurd image (I downloaded iso yesterday) and up to date, and I’m up to date in netBSD. Your commit is applied, and I use the same command to build than yours <Pellescours>I’m checking if it’s not a library which is missing on my image <Pellescours>damo22: ok, I installed some missing library and removed build artifacts and it build <akib>How much effort will be needed to port Hurd to another microkernel? And what features the microkernel needs to provide? <Pellescours>akib: I know some effort was made to make hurd running with L4 but I think that Hurd on gnumach is the only one that is still moving <akib>oh! i missed that. thanks! <akib>and what does the hurd expects from the host microkernel? <youpi>essentially, memory management, thread scheduling, and IPC <youpi>currently mach also provides disk drivers, but that's being removed <akib>does hurd contains disk drivers? will hurd work if i port it to a microkernel without those drivers? <youpi>disk drivers will come from rump <youpi>that's currently done, see IRC logs, damo22 is on it <akib>what is rump? i found an wikipedia article on rump kernels, is this what you mean? <youpi>see the rump kernel page on the hurd wiki <akib>is debian hurd free (as in freedom)? <akib>it is not listed on free gnu/linux distro list <akib>oops! typo. i mean debian gnu/linux is not fully free. is debian gnu/hurd fully free? <youpi>debian gnu/linux is fully free <youpi>that is, the real Debian gnu/linux <youpi>the contrib and non-free sections are not part of debian <youpi>it happens that contrib and non-free are not compiled for gnu/hurd <akib>and what about those non-free blobs in linux kernel? <youpi>they're not on official debian cd images <akib>then what's the kernel of debian gnu/linux? linux-libre or debian's own patched kernel? <youpi>ah you mean some in-kernel-source blobs <youpi>I don't know the status of this in linux, but debian is not supposed to ship them <youpi>if there are some of them it's a bug, just report it <youpi>note that I really mean the *blobs* <akib>what's the glibc and libpthread repo in hurd savannah project? <youpi>it's not there it's in upstream glibc <akib>if i try to port hurd to another microkernel, should i grab the source from those repo or from the official repos? <akib>is libpthread is merged into glibc? <akib>looks like glibc source code is huge <akib>wow, it's more than a million lines! <akib>i have no idea where to start hacking on glibc and pthread <akib>except that i have to reimplement mach/ and sysdeps/mach/ <trollorozzo>akib: why you want to port the Hurd to another microkernel? Which micro kernel? The Minix 3 one? <trollorozzo>Does gnumach automatically schedule the receiving task when one send a msg from say task A to task B? <Gooberpatrol66>The easiest way to port to another microkernel to start with is to write a compatiblity layer for mach, right? So one would implement all the stuff listed in the gnumach reference manual? <gnu_srs>Howdy, Maybe the best effort for now would be to make Debian GNU/Hurd a releasable architecture? <gnu_srs>Contributions are extremely welcome: First issue is to finish/complete the rumpdisk port for Hurd! <gnu_srs>Then of course USB and audio will follow :) <ThinkT510>I know it is massive work but is SMP on the roadmap or has it been firmly decided that it is not worth implementing? <ThinkT510>I was under the impression it required major changes to mach <youpi>mach already had smp support <youpi>it has "just" been untested for a couple of decades <ThinkT510>It really is nice to see hurd continue to develop